I’ve been playing round with getting good quality compressed speech with LAME. (See here for iTunes.)Â Here’s what I’ve found.
1. SMALLEST FILE SIZE ACCEPTABLE QUALITY.
For very small file size (3MB for 20 minutes) and acceptable quality,Â use the LAME encoder with the following arguments in the ‘Encoding options’ Box:
–abr 16 -q 0 Â -m m Â (gives very small file size but some artefacts audible)
Here are some better options, note that the higher bitrate does not always sound better! I think a lower bitrate decreases the frequency of the lows pass filter so in some cases lower bitrate sounds better as it takes of the harsh tops, depends on the person and how their voice sounds.
–abr 24 -q 0 Â -m mÂ (gives small file size but you can only just tell it’s compressed)
–abr 32 -q 0 Â -m m
–abr 48 -q 0 Â -m m
abr 16 means a bitrate of 16kbps
-q 0 means best quality available at that bitrate
-m m means mono
You can hear some low pass filtering on the above speech, the highs are missing, but this can make for a better listening experience with spoken word, it can be less harsh.
3. BETTER QUALITY AND ACCEPTABLE FILE SIZE.
For about twice the size files (8MB for 20 minutes of speech) but very very good quality try this in LAME encoder:
-V 8 –vbr-new -h -q 0
4. BEST QUALITY. Use iTunes. Although LAME is better for music, and lower bitrate spoken word, iTunesÂ is great for 40kbps spoken word and above. CLick here to read about the best settings on importing spoken word into iTunesÂ here.